JPS60128559A - Kana (japanese syllabary) kanji (chinese character) converter - Google Patents

Abstract

PURPOSE:To improvae the efficiency of generation of a document by displaying and outputting the result of plural converting candidate units to a long input character string such as so-called solid sentence in an easily understandable way. CONSTITUTION:After the 1st character string representing the reading and inputted via an input device 1 comprising a Kana character reader or the like is converted into, e.g., Kana character code, the string is given to Kana/Kanji converting section 2 comprising a block extracting section 2a, a total paragraph series extracting section 2b, a paragraph extracting section 2c and a converting dictionary 3. Then the Kana/Kanji converting section 2 transfers the 2nd character string comprising Kana/Kanji mixed representation obtained to the paragraph series to an output data memory 5 of an output control section 4, which converts the data into a prescribed display output from and outputs the result to a display device 7 via a document display memory 6. Thus, an object Kana/Kanji converting charactr is selected simply from the conversion candidate characters of the homonym and homo-paragraph so as to form efficiently a Japanese sentence.

Description

Detailed Description of the Invention (Technical Field of the Invention) The present invention converts a long sequence of continuous kana characters, such as those input as sentences, into kana-to-kanji characters while appropriately dividing them into phrases to convert them into kanji and kana characters. The present invention relates to a kana-kanji conversion device that can effectively create Japanese δn sentences.

[Technical background of the invention and its problems]

In conventional Japanese word processors, the unit of Kana-Kanji Jin-no-Ko is generally limited to a single phrase, and in most cases, the input of noun compound words is limited to several units at most. . Therefore, when inputting and creating Japanese sentences using this type of device, the operator must always be aware of the units of words or phrases, which places a heavy burden on the operator. Therefore, recently, various studies have been attempted to input a sequence of pronunciations of kana in sentence units, so-called solid sentences, without restricting the units of text input, and to perform kana-kanji conversion processing on the solid sentences. Specifically, they achieve their purpose by, for example, recursively performing clause analysis processing. However, there are problems in that this requires a considerably long processing time and consumes a large amount of buffer memory. Furthermore, although it has been considered to simplify the algorithm for phrase analysis processing by limiting the processing time and memory amount, it cannot be denied that the accuracy of the conversion processing deteriorates. However, a major problem has been how to display and output the conversion results obtained in this way so that the instruction for selecting the homophone a can be easily handled.

For example, if the character string ``sandaka wo motoru'' is manually written in kana, it is expected that it will be possible to mechanically divide the phrase into phrases ``sandaka wo,・'motoru'', or it can be divided into phrases ``sandaka wo mo/stop''. Divided forms are also grammatically possible.

In this case, it is generally considered that the so-called longest match is most likely from a historical perspective, but based on this empirical rule, always perform phrase analysis starting from the beginning of the input character string i). Extracting only "also/stop" will significantly degrade the conversion accuracy. Therefore, it becomes necessary to extract a plurality of conversion candidates, such as ``also/stop the balance'' and ``to request the balance,'' and to leave the final decision to the decision of the operator A.

Also, for the entry ``Keisanshiki'', if 7i, ``calculation formula'', is not registered in the dictionary in the device, the result will be r Ct 07 formula in the same way.''
, ``compute/tree'' produces multiple transformation results. Furthermore, conversion results such as ``hair/hentai,/shiki'' also occur. Furthermore, even if [Rukei-“'Exile’].

If the word `` is registered in the dictionary, candidates such as ``Claim the balance/Exile/Calculation formula'' will also appear.

By the way, as an evaluation process for outputting the most probable one in the first order among various conversion results for such an input, for example, priority is determined in order of the number of clauses or words that make up the whole. A method is being considered. Specifically, for example, in response to the input of [Kou Gaku Shiyoto (), you can input "high amount/income" as [section ga/phrase/place 1q] or "
It is determined that the term is more likely than ``/kudokoroto/season''. In this case, it is natural and preferable to output homophones with priority in the order of frequency of use. However, even if the number of phrases that make up the whole phrase is the same, the way they are separated may be different, and when outputting analysis results that include a large number of phrases in order to reduce omissions in conversion, it is necessary to handle them. Data structure becomes complicated. Moreover, the problem is how to display and output the plurality of conversion results in order to make it easier for the operator to perform selection operations.

For example, if only the homophonic parts within each phrase are shown with elegance such as brightness changes, it is not possible to tell whether there are other phrase series with different phrase breaks. Furthermore,
There were many problems, as it was necessary to select the homophone dissent clause Sugizo in the selection pond for the homophone objection edict.

[Object of the Invention] The present invention has been made in consideration of the above circumstances, and its purpose is to obtain results for a plurality of conversion candidate units for a long input character string, such as the so-called 9 sentences. An object of the present invention is to provide a kana-kanji conversion device that can display and output information in an easy-to-understand manner and improve the efficiency of document creation.

[Summary of the invention]

The present invention uses a dictionary search unit in which a plurality of words are registered in a dictionary to extract a series of clause units from a series of input character strings,
When converting and outputting each of these clauses into the kana-kanji mixed notation corresponding to the pronunciation of the clause, the earliest clause starting point among the mutually different break positions of the plurality of clause series for the input string is determined. As a reference point, the free-standing HD part of the next clause that exists after this reference point and the bowl 1
Kana-kanji conversion is performed using the adjunct part of the previous clause separated by M points as a unit of conversion candidates, and the conversion candidates are displayed and output in order.

〔Effect of the invention〕

Thus, according to the present invention, for example, in response to the above-mentioned input of ``Keisanshiki to obtain Zandaka'', phrase series candidates such as ``Also/stop the balance'' and ``Claim the balance'' are created. Click [Balance] to select the conversion candidate result.
Since these messages are displayed and output in order, such as "stop request" or "request [balance]", the process for selecting the next candidate with the same sound can be greatly simplified. That is, for example, regarding the part [Modome] mentioned above, the word "Motome" does not actually exist, but other homophone word information and separate table arrangement exist in the pronunciation "Moto" part. In other words, it is shown that there are other candidates corresponding to the same pronunciation, and in the same way as selecting homophones in a bunsetsu, the command to switch the next homophone candidate simplifies the selection operation of differently structured bunsetsu strings. Moreover, it can be performed efficiently. Therefore, without providing a special instruction key to select another candidate in a string of different structure phrases, it becomes possible to efficiently perform the selection process using the conventional homophone next candidate key. A tremendous effect can be produced.

[Embodiments of the invention]

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of an embodiment device. The input device 1 includes a keyboard device, a voice recognition device, a kana character reading device, and the like. A character string (first character string) representing the pronunciation inputted via the input device 1 is converted into, for example, a kana character code, and then provided to the kana-kanji converter 2 . In addition, the first character string representing the above reading may be, for example, hiragana, katakana, katakana,
It is indicated in Roman letters, etc.

The kana-kanji converter 2 is, for example, a block extractor 2a.

It is composed of a total phrase series extraction section 2b, a phrase extraction section 2c, and a conversion dictionary 3. The kana-kanji conversion unit 2 generates a second character string consisting of a display character string mixed with kanji corresponding to the first character string transferred from the input device 1, and outputs this to the output control unit 4. is giving to The block extraction unit 2a is provided to efficiently analyze particularly long input character strings with respect to the first character string, and extracts a preset value N1, for example, N=4. , contains the phrase analysis results to be associated as a phrase series within N sentences WJ, and when there is no corresponding phrase analysis result,
The first input character string is divided into several block sections. Then, the pronunciation kana string of each block divided in this way is sent to the total bunsetsu series extraction section 2b, and the kana-kanji conversion result for each block found by the total bunsetsu series extraction section 2b, that is, the second character The columns are sequentially sent to the output control section 4.

The total clause series extracting unit 2b divides the first character string into divisible clause series using the clause extracting unit 2c, and for each of these clause series, the corresponding kana-kanji mixed notation is written. The conversion result is output to the block extraction section 2a. The phrase series is used for all combinations of the input character string that can be divided into phrases, and after ranking the most likely candidates based on, for example, a priority evaluation, the first phrase is selected. are output in order. In addition, when evaluating the priority, in general, there is a higher tendency for phrases with a small number of phrases to correspond to sentences containing kanji that are the input target, so for example, phrase sequences with a small number of phrases are given priority for output. will be carried out.

The phrase extraction unit 2C performs a collation search between the input character code string and character strings (words) registered in advance in the conversion dictionary 3, and extracts written characters containing kanji that correspond to the first character string. We are looking for a second string of characters. The conversion dictionary 3 has an input heading table area 3a, an output heading table area 3b, and a part of speech area 3C, as shown in FIG.
A first character string representing the reading is stored in the input heading guide jg3a, and a second character string consisting of written characters mixed with kanji corresponding to the first character string is stored in the output heading table area. 3b.

The part-of-speech area 3C stores part-of-speech information for the first and second character strings.

Therefore, the phrase extraction unit 2C uses the conversion dictionary 3 for the given input character string, for example, using the known longest forward match method.
The character string (
The system searches for a word), analyzes its inflected endings and adjuncts, and selects the part of the input string that matches the length of the input string from the beginning as the clause extraction result. At this time, the above-mentioned analysis of the conjugated ending is performed based on the part-of-speech item stored in the part-of-speech area 3C. Then, a second character string consisting of notation characters including kanji characters corresponding to the phrase extraction result found by this analysis is read out from the output heading table area 3b and output. Furthermore, at this time, the total phrase series extraction unit 21) inputs the input unit (phrase extraction result) to the phrase extraction unit 2C into various combinations that can be divided into phrases in the block section determined for the input character string. , and found the most probable bunsetsu sequence among them.

The kana-kanji conversion unit 2 transfers a second character string composed of kana-kanji mixed notation determined for such a phrase series to the output data memory 5 of the output control unit 4. The output control unit 4 converts these data into a predetermined display output format,
It is output to the display device 7 via the document display memory 6.

As shown in FIG. 3(a), the output data memory 5 includes a combination table 5a, a mapping table 5b, and a heading type table 5C. It is stored together with the combination structure. This example represents the data storage structure of the example of the Japanese sentence shown in FIG. is described. The rows of this table 5a describe the interpretational types of bunsetsu structures, and the columns describe the connections of clauses in each bunsetsu structure in order. That is, in the first block (1! Complement or 1 type,
It is shown that there are two types of sequence candidates for the second block and three types of sequence candidates for the third block. Further, each of these numerical values indicates an element number of the mapping table 5b. In addition, the headwords of homophone dissonance words that exist for each phrase unit are grouped and stored in table 5C, and each group in headline class table 5c is described using each element of mapping table 5b as a pointer. In this way, the correspondence with the input reading kana position is also described at the same time.

FIG. 3(b) shows the contents of the output data memory 6.
This is a table structure of the document display memory 6 that stores character display information to be output to iN@7. This memory 6
is based on the contents of the output data memory 5, and compares and matches homophone meaning words and different structure clause strings located at the same pronunciation kana position, and first, the above-mentioned at a common clause break point among all clause series candidates for the input character string. The input character string is divided, and the common appendage character part and the only passable conversion result part are separated from the other parts and described. In addition, for conversion candidate results that have the same reading kana in the attached word part for the headword character at the beginning of a certain clause,
It is divided according to the beginning of the previous clause, and the character string of the adjunct after the division is joined to the next clause. In other words, even if it is a character of an adjunct word in a clause of a phrase series, if the corresponding part with the same reading is part of the beginning of an independent word in another phrase series, the character of the adjunct word is Conversion of the next clause (described by combining with candidates.As will be described later, these homophones are displayed and output with their display attributes different from the normal display attributes.Figure 5 (a) ) to (d) are display examples, and the shaded areas indicate different display attributes.Changes in display attributes are performed, for example, by inverting displayed characters, blinking, changing brightness, underlining, etc.

Next, the kana-kanji conversion process in the kana-kanji conversion section 2 will be explained using the example of kana character input shown in FIG.

The block extraction unit 2a extracts a sequence of up to N phrases from the beginning of the input character sequence as much as possible. Here, if N is set to 4, for example, in the example shown in Figure 4(a), the entire input sequence is first panned to the clause extractor 2c], and by the longest match method, ``mo shite'' is set as the first clause result. obtain. Next, the longest matching result is found in the same way, with the next bunsetsu starting character position starting after this bunsetsu break, to obtain the bunsetsu ``kondonoto.'' Such processing is repeated in order to obtain the first clause series candidate ``And/Kongonoto/Ushiha/Kaiteishi'' as shown in item ``A'' in FIG. 4(b). Next, in order to 1q the bunsetsu series obtained for this item "A" and another bunsetsu series, the last character of the third clause result "Ushiha", that is, is analyzed as an adjunct in that clause. Delete "wa" and send it to the phrase extraction unit 2C, and in the same way, obtain the longest-matching phrase "ushi", followed by "ha" as a phrase starting with "ha".
I put it on. Thereafter, in the same manner, each time 19 short phrases are created by the third, second, and first phrases, another string of phrases following these phrases is sequentially watered. In this way, all four-clause sequences that can be divided into phrases for the input character string are divided into Figure 4 (b
). At this time, a corresponding heading edict candidate (second character string) containing kanji is also noted at the same time.

Next, among these sequences, attention is focused only on the candidate whose total length is the longest (the bunsetsu sequence). This is based on the fact that, as shown above, the sequence with the smallest number of clauses in the input is more likely to contain the conversion result intended for input. This means that it is sufficient that the number of clauses in one block is the minimum, and that if the number of clauses is the same, the block has a longer length.

Among the results shown in FIG. 4(b), the longest phrase series are those shown in the item "a" and the item "tsu". Then, find the next (Q) where these phrase series have a common phrase break. In this example, only [and/~] and [~wa/~] are used as 5 as common text delimiters. The block extraction unit 2a determines these two positions as block divisions, and defines the first block section as "and" and the second block section as "Kongo no Toshiwa." Then, the character strings in these sections are sequentially analyzed by the total phrase string analysis section 21), and the conversion results are sent to the output control section 4 (A).

As a result, the first block section becomes a candidate for only "and", and this information is first sent to the output control 034. In this case, since there are no other homophones, the text is displayed as character data (conversion result) in the document in its normal form. After that, the second block section is analyzed.

Here, the total bunsetsu string analysis unit 2b is to use a round-robin method to find character sequences corresponding to the pronunciations of a given block section, but in reality, as shown in Figure M4 (b), the bunsetsu sequence Therefore, it is sufficient to select only those that correspond to the range of the specified interval. As a priority evaluation, for example, if only the candidates with the minimum number of clauses are selected, the analysis result will be as shown in the item "π" in FIG. 4(d). Of course, other clause candidate sequences may also be provided to the output data memory 5. For example, it is also possible to add "What are the 7 magazines to ask about in the future?" to the output results.

Next, the block extraction unit 2a extracts blocks in the same way for the remaining part of the input character sequence for which blocks have not yet been determined, that is, the character string [Kaiteishige...]. Find the unit and enter the conversion result as shown in item rI[IJ in FIG. 4(d). Subsequently, the conversion result "Limited" is set as the item r IV J in FIG. 3(d), and the conversion process for the entire input series is completed.

The above conversion results are sent to the output control section 4 for each block. The output/control unit 4 converts each block of data stored in the output data memory 5, sequentially stacks it in the document display memory 6, and outputs it to the display device 7. That is, the output control section 4T: (prepares document display data as shown in FIG. 3(b), and first outputs it to the display device 7 as shown in FIG. 4(a). In the example, for the bunsetsu candidate ``Shiki no to'', the independent word part is separated as ``Shikoku °'', and the attached in part is separated as Pa no and ``'.The same applies to the bunsetsu candidate ``Ushiwa''. “Cow” and “” are separated into “”.

Furthermore, in the second clause series candidate, the independent word part is separated as ``from now on'', the adjunct part as ``pa'', and ``investment wa'' is separated into ``investment'' and ``wa''. Therefore, in this case, ``'°'' and ``wa'' are both common appendage characters, and since there is no other homophone for ``hereafter'', these character parts are displayed in the normal display form. Furthermore, since there are a plurality of conversion candidates (homophones and allographs) for other characters, they are displayed with high brightness, for example, to emphasize these candidates. In other words, although `` and '' in the phrase ``Koku no To'' are added RHE, they are treated together with the next phrase word ``ushi''.

Therefore, the display device 7 shows [and future [and Ushiko is [
``Limited to limit valves'' will now be displayed. Note that characters in brackets [ ] indicate characters displayed with high brightness.

In order to select the desired headword here, the input device 1 is provided with a selection key as shown at 18 in FIG. 1, for example. This selection key 1a has the role of changing the display to the next candidate when the cursor 1b is placed on the [and cow] part as shown in FIG. 5, and the selection key 1a is operated in this state. It is the responsibility of Therefore, in this case, instead of the 4% construction in the first row of table 5a shown in FIG. The output is displayed as shown in FIG. 5(b).

Then, as the selection key 1a is further operated, "investment" will be changed to other homophones one after another, such as "fighting spirit" and "clairvoyance", and will return to the original display of "and cow". . still,
These operations are performed based on data stored in the document display memory 6 shown in FIG. 3(b).

By the way, the document correspondence table 6a shown in FIG. 3(b) represents the correspondence between the shock values 11 to i7 on the display screen of the display device 7 and output data (conversion results).

The display word table 61+ shows the contents of the output data. However, in displaying the conversion candidates, first, "■ and cow" in the display word table 6b is designated by data 13 in the document correspondence table 6a, and this is displayed. Thereafter, when the selection key 1a is operated, the pointer is advanced to the next candidate "Investment" in that block according to the data 13; J5, and the display is switched. In this way, when the pointer is advanced and the display is switched to "■ Freeze to death", the display is controlled to return to "■ and cow".

As shown in this display word table 6b, the multiple conversion candidate results analyzed based on the bunsetsu unit as described above are organized according to homophone and allograph relationships, and the unit of separation is partially changed. .

Similarly, for the character string "Kaitishige...", the independent word part is treated as "revised", the suffix a word part is treated as "'shi", and the "t" part is treated as "resource". Because there is a phrase word, this "stake" is output by combining it with the next phrase pyramid. Since there is no break between the words that all candidates have in common until the following "Hatsu ni", these are output all at once. Note that in machine processing, each independent word is treated as a clause.

6 to 9 show the flow of the above-mentioned processing of this apparatus. In the control flow shown in FIG. 6, the input key codes obtained from the input device 1 are constantly checked, and if the input codes are kana character codes corresponding to the pronunciation of a Japanese sentence, they are sequentially stored in the stack. If the input code indicates a conversion request, the kana/kanji conversion process shown in FIG. 7 is performed. If the input device has a conversion request key, the conversion request is generated by the operator inputting a character string of an appropriate length and then pressing the conversion request key. Further, regardless of whether the input device has a conversion request key or not, it is desirable to automatically generate the conversion request when it is detected that the input character is a code indicating a punctuation mark, for example.

If the input code corresponds to the selection key 1a, the homophone selection process shown in FIG. Yoto processing is performed on the Further, FIG. 9 shows the editing and output processing of the conversion candidates in FIG. 7 according to the present invention.

Generally speaking, word processors are known to perform the selection of homophones) for each conversion result sequentially, and to select them all at once after inputting, for example, one page of character strings.
In the device of the present invention, either of these methods may be used.

As explained above, according to this device, when a relatively long input kana character string is analyzed and converted into a character string containing kana and kanji to create a Japanese sentence, a large number of characters are generated as a result. The phrase string candidates can be displayed simply and clearly, and the efficiency of the operator's homophone selection operation can be improved. In other words, it is possible to select homophones not only for one phrase candidate, but also for candidates in a series of different phrases with different phrase breaks as described above. Therefore, it is possible to easily select the target kana-kanji conversion characters from the conversion candidate characters of homophone 11B and homophone different phrases, and to create a Japanese sentence extremely efficiently. Furthermore, the burden on the operator can be significantly reduced, and its practical advantages are enormous.

Note that the present invention is not limited to the above embodiments. For example, instead of displaying homophones for display output sequentially in the document, the homonym group Shushun is displayed below the text,
This may be selected by the operator. Furthermore, when the desired character is not selected from the homophone group, subsequent conversion results may not be displayed to prompt the user to select it. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of drawings]

FIG. 1 is a schematic configuration diagram of an apparatus showing one embodiment of the present invention, and FIG.
Figure 3 shows the configuration of the memory of the conversion dictionary, Figure 3 shows the configuration of the output data memory and sentence @ display memory, Figure 4 shows the input character string and its conversion process, Figure 5 is converted!
Return? FIGS. 6 to 9 are diagrams showing an example of a display of tti, and FIGS. 6 to 9 are diagrams showing an example of a control flow of conversion processing. 1... Input device, 2... Kana-kanji converter, 3...
Conversion dictionary, 4... Output control unit, 5... Output data memory, 6... Control table, 7... Display device, 1a
. . . first ill plain key 2a . . . block extractor,
2b... Total clause series extraction unit, 2C... Clause extraction unit,
3a...Input heading table area, 3b...Output heading table area, 6a...Intra-sentence correspondence table, Gb...Display word table. Applicant's agent Patent attorney Takehiko Suzue Figure 3 (b) 4i2! (d) No. 50 No. 7 Fig. 81E No. 9 J

Claims (3)

[Claims] (1) An input device for obtaining a series of input character strings, a dictionary search unit in which a plurality of words are registered in the dictionary, and phrase sequence extraction for extracting a sequence of phrase units from the input character string using the dictionary search unit. and a result output unit that converts each clause found by the clause series extraction unit into a kana-kanji mixed notation corresponding to the pronunciation of the clause and outputs the result output unit. When the series extraction unit obtains multiple clause series for an input character string, the earliest clause start point among the mutually different break positions of these clause series is set as a reference point, and the clauses that exist after this reference point are A kana-kanji conversion device characterized in that the independent word part of the next clause is combined with the adjunct part of the previous clause separated by the reference point to perform kana-kanji conversion as a unit of one conversion candidate. (2) The result output unit displays and outputs the units of conversion candidates after the reference point by changing the display attributes by reversing the displayed characters, blinking, changing the fi degree, or adding underlining. The kana-to-kanji conversion device according to item 1. (3) The display results of conversion candidate units after the reference point are to output the kana-kanji conversion results of multiple conversion candidate units having a common reference point in order in response to a request to switch to the next homophone candidate. A kana-to-kanji conversion device according to claim 1.